Could have done otherwise, action sentences and anaphora

Could have done otherwise, action sentences and anaphora HELEN STEWARD What does it mean to say of a certain agent, S, that he or she could have done otherwise? Clearly, it means nothing at all, unless the anaphoric devices within the sentence have been anchored to definite antecedents. In this paper, I shall argue that there may be more ways of effecting this anchoring than is commonly supposed, and hence more questions potentially available to be asked by means of the formulation Could S have done otherwise? than is generally assumed to be the case in most of the relevant literature. What are the anaphoric devices whose reference needs settling before we can make an assessment of any particular S could have done otherwise claim? Clearly, one such device relates to tense; could have introduces a past-tense morpheme which requires some specification by an antecedent sentence of what Reichenbach calls a point of reference (1947: 287 98). Temporal anaphora of this sort are interesting and complex, and have already been the subject of some scrutiny in the literature relating to could have done otherwise (see e.g. Wiggins 1991: 273 74 and Chisholm 1967). I will not however say more about them here; my concern will rather be with the functioning of the verb phrase do otherwise, which has received less attention. What do otherwise appears to demand for its sensible interpretation in the context under consideration is an antecedent sentence of the form S ϕ-ed, where ϕ is a verb of action. Given an antecedent sentence of this sort, it is natural to suppose as is standardly assumed in the literature that what it means to say that S could have done otherwise is that S could, at the point of reference in question, have done other than ϕ. I do not want to deny that this is one of the things it is possible to mean by saying that S could have done otherwise ; nor even that it is certainly what is usually meant. But I do want to suggest that another interpretation is also available, an interpretation which can be of considerable importance in some of the philosophical contexts in which claims of the form S could have done otherwise are made and assessed. It may be helpful, before I attempt to describe how this other possibility can arise, to consider another anaphoric device which very clearly makes possible an ambiguity parallel to the one I wish to suggest exists ANALYSIS 66.2, April 2006, pp. 000 000. Helen Steward

also in the case of otherwise. Consider the phrase something else as it occurs in (1) Jim gave Tina a book for her birthday. She d have preferred something else. Something else might mean here something other than a book, but I think it is clear that it might also mean something other than the book he in fact gave her it would be perfectly permissible to continue: She doesn t much like novels. Non-fiction would have been a better idea. Something else, in other words, though its antecedent here is a quantificational expression ( a book ) is able to function, in effect, as though it were reliant for its explication on what Gareth Evans calls an E- type pronoun one which appears to be able to refer to something particular, even though its antecedent is some kind of quantificational phrase (Evans 1985). To take one of Evans own examples, in the sentence Mary owns a donkey, and John beats it, the pronoun it seems to manage to refer to what Evans calls the verifier of the antecedent quantificational expression the donkey which Mary in fact owns. In much the same way, I suggest, an anaphoric device like something else, when its antecedent is a quantifier (as with a book ), may in effect work as though it meant something other than that thing, where that thing is whatever verifies the antecedent. It need not mean (though it may mean) something of another kind. Turning now to otherwise, my suggestion is that a similar ambiguity may arise in the context of the claim, made subsequent to a sentence of the form S ϕ-ed, that someone could have done otherwise. Consider the following example: (2) Brutus stabbed Caesar. He could have done otherwise. It is natural to assume that what is meant here is that Brutus could (at the point of reference in question) have done other than stab Caesar. But suppose we take seriously the suggestion that action sentences quantify implicitly over events. 1 In that case, the claim Brutus stabbed Caesar is semantically more complex than might at first have been imagined. Suppose, for instance, we accepted that it had something like the following underlying semantic structure: 1 Clearly, here is not the place to defend the claim that action sentences do indeed have a semantic structure of this kind; although I believe the case for thinking so is very strong. My argument should therefore really be interpreted as having the conditional conclusion that if action sentences are quantifications over events, then the ambiguity I am about to discuss will arise.

x [Stabbing (x) & Agent (Brutus, x) & Object (Caesar, x) & Culminate (x, before now)]. 2 What might now be made of the subsequent claim that Brutus could have done otherwise? Obviously, there is a difficulty I am currently unsure whether very superficial, or very deep 3 about interpreting the sentence: Brutus could have done otherwise at all, once one imagines it preceded by a sentence whose form is dominated by an existential quantification of this kind. For once we rewrite the first sentence in this way, it no longer even delivers an antecedent of the sort that can makes sense of a subsequent claim to the effect that the agent of the action could have done otherwise. For that, it seems, we need some part of the antecedent quantificational structure to maintain a subject-verb form on which the general and schematic verb do in the subsequent sentence can pick up. For this reason, and indeed for the more general one that it seems slightly more perspicuous, I propose a minor modification to the Parsons formulation. Parsons s treatment of action sentences exploits what he calls the Agent relation the relation in which an agent stands to any token action she performs. I suggest that instead of the Agent relation, we utilize the Performs relation which is definitionally interchangeable with the Agent relation in the following way: S Performs e = df S is the Agent of e Because of the definitional interchangeability here, the divergence from Parsons is trivial; and yet the alteration delivers what we are after, which is a conjunct within the scope of our quantificational expression which directly relates the agent to her action, and which therefore is available to connect with the verb structure of the subsequent could have done otherwise claim, in such a way that a cross-reference can be effected. Our analysis, then, has now become: x [Stabbing (x) & Object (Caesar, x) & Culminate (x, before now) & Performs (Brutus, x)] 2 There are of course many different versions of the proposal that quantificational structure can be discerned in action sentences. I take this particular one from Parsons 1990, but the precise details are not important. So far as I can see, my point holds for any view on which it is maintained that action sentences are quantifications over events. 3 One reason for thinking it may be deep: one might, even while accepting Frank Ramsey s claim that action sentences contain concealed existential quantification (Ramsey 1978: 43), feel nevertheless that no formulation which dethrones the agent from the dominant position in the sentence can adequately represent an action sentence. See Wiggins 1985 for an alternative way of incorporating the existential generalization.

What I would like to suggest is that this hidden semantic structure makes it possible for a subsequent claim to the effect that Brutus could have done otherwise to mean that Brutus could have done other than perform the particular action that verifies the existential quantification 4 just as the claim that Tina would have preferred something else can mean that she would have preferred something other than the particular book that verifies the claim that Jim gave her a book for her birthday. It can mean, in other words (though it need not), that Brutus could have refrained, not from stabbing Caesar, but rather from performing the particular stabbing that he did in fact perform. One might press the question whether this reading is really available. One legitimate question might be whether an anaphoric device can really function in such a way as to pick up on logical structure which is largely concealed by the surface form of a sentence. But there are strong reasons for thinking that there can be no general impossibility of this sort. It has been argued, for example, by Jim Higginbotham (1983) that the grammar of certain types of perceptual report (e.g. John saw Brutus stab Caesar ) suggests quantification over underlying events, and it seems plausible that anaphoric devices may be able to pick up on the verifiers of such quantifications. Consider: John saw Brutus stab Caesar. Did you see it, too? Here, the pronoun it appears to refer to an event the stabbing of Caesar by Brutus which is the verifier of an implicit quantification in the antecedent sentence. And if this kind of phenomenon is possible in general, there seems no particular reason why it should not arise also in connection with the functioning of a device like do otherwise. Another kind of scepticism might derive from the fact that it is very hard to imagine a situation in which there is any point in making the distinction upon which the ambiguity I am alleging exists depends. It might be said, with some justice, that the parallel I have sought to draw with (1) above is vitiated by the fact that, although there is nothing easier than to imagine an ordinary situation in which, though it is not true that Tina would have preferred something other than a book, it is true that she d have preferred something other than the book she was in fact given, it is very difficult to imagine one in which, though Brutus could not (at the relevant time) have done other than stab Caesar, he could have done other than perform the very stabbing he did in fact perform. Thus, while it is easy to illustrate and understand the ambiguity in (1), it is very hard to see how anything comparable could arise with respect to (2) so that the distinction I have attempted to draw might seem to collapse for sheer want of purpose. I concede that the ambiguity in which I am interested is 4 What if there is more than one such verifier? My assumption is that context normally selects the relevant action just as it selects the relevant book in (1).

unlikely to surface very often, if at all, in most ordinary contexts. But there are philosophical contexts, I think, in which it can, and does, make its presence felt. Harry Frankfurt has challenged the principle that a person is morally responsible for what she has done only if she could have done otherwise (the Principle of Alternate Possibilities) by describing certain apparent counterexamples, in which we are invited to have the intuition that an agent is morally responsible for what she has done even though she could not have done otherwise (Frankfurt 1969). For instance, we are asked to imagine that Black, an agent with a number of unusual powers, 5 wants Jones to do something shoot the President, say but that he does not want to take action to ensure that Jones will act as he wishes unless this is really necessary. He decides, therefore, to wait until Jones is about to make up his mind what to do, and resolves to intervene to ensure the outcome he wants only if it becomes apparent that Jones is not going to choose to do what Black wants him to do. Should this happen, Black will take effective steps to see that Jones does indeed decide to do, and then does, what he wants him to do. But fortunately, as it turns out, Black does not need to intervene at all; Jones decides for himself to shoot the President in any case. Frankfurt argues that in a case such as this, Jones is morally responsible for shooting the President, even though Black s intended counterfactual interventions mean that Jones could not have done otherwise. If he is right, the Principle of Alternate Possibilities must be false. But many philosophers have retained an uneasy conviction that there is a clear sense in which it remains true to say, under these conditions, that Jones could have done otherwise, though it has not proved very easy to say what this conviction amounts to, or to defend it against the obvious truth of the claim that Jones could not, under the conditions imagined by Frankfurt, have done other than shoot the President. What might these refuseniks be thinking? Some have resorted to the natural idea that more explicitness with regard to times will help and have argued e.g. that Jones could, at any rate, have done other than ϕ at t (where t is the actual time of the action) only to be met with ever more ingenious counterexamples which make it impossible for Jones even to avoid ϕ-ing at t, though he remains intuitively morally responsible for what he does (see e.g. Mele and Robb, 1998). My suggestion offers what I think is a more robust and 5 Frankfurt says that Black can be allowed whatever mind-reading powers are required to enable him to decide correctly whether and when he should act, and also whatever coercive means are deemed necessary to guarantee that Jones really cannot do otherwise in the situation concerned threats, potions, hypnosis, the surgical tools necessary to produce the requisite changes in Jones s nervous system and the expertise to use them, or whatever.

promising explanation of the persistence of the conviction that Jones could have done otherwise, even under the conditions imposed by a Frankfurt-style case. On my account, that conviction is dependent upon the perception that Jones could, at any rate, have done other than perform the very shooting he in fact performs, and on the fact that this is one thing that might conceivably be conveyed by the claim that Jones could have done otherwise. This suggestion offers a more robust response to Frankfurt than does the strategy that resorts, rather, to times, because of the overwhelming plausibility of the claim that no possible world in which it is Black, rather than Jones, who initiates Jones s ϕ-ing, is a possible world in which the same individual action occurs as occurs in the actual world. 6 And moreover, if I am correct, it would seem to be true not only that there is a sense in which Jones could indeed have done otherwise in the Frankfurt scenarios, but also that there is a reading of the Principle of Alternate Possibilities itself that remains unscathed by Frankfurt-style cases the reading on which what it states is that an agent is morally responsible for what she has done only if she could have done other than perform the very action she did in fact perform. It might perhaps be objected that even if it is true that one may sometimes read an individual could have done otherwise claim in the way I have suggested, the phrase cannot be read this way as it figures in the Principle of Alternate Possibilities. For that principle states that a person is morally responsible for what he has done only if he could have done otherwise, and what a person has done, it might be argued, is always a kind of thing, never a particular action, so that the interpretation of the could have done otherwise claim which I have highlighted is unavailable in this context. It seems to me, however, that the same ambiguity is present in the reference back to the general formulation what he has done as exists in a sentence of the form S ϕ-ed. What has S done? Well, S has ϕ-ed. But S has also performed the very ϕ-ing that she in fact performed. So there remains a reading on which PAP still seems defensible: a person is morally responsible for ϕ-ing on some particular occasion only if she could have done other than perform the very ϕ-ing that she in fact performed. Others have of course suggested that there are principles relating moral responsibility to certain sorts of alternate possibilities, which do not fall 6 This claim needs more defence than I am really able to give it here; suffice it to say that it seems to me that something must have gone badly wrong with a view of action on which it is possible for one agent to initiate and control what are nevertheless the genuine actions of another. For further defence see Widerker 1995.

to Frankfurt-style examples. 7 The present suggestion is distinctive, though, in that it proposes that PAP itself admits of a legitimate interpretation which withstands Frankfurt s attack. References Balliol College Oxford OX1 3BJ, UK helen.steward@balliol.ox.ac.uk Chisholm, R. 1967. He could have done otherwise. Journal of Philosophy 64: 409 17. Evans, G. 1985. Pronouns, quantifiers and relative clauses (I). In his Collected Papers, 76 152. Oxford: Oxford University Press. Frankfurt, H. 1969. Alternate possibilities and moral responsibility. Journal of Philosophy 66: 829 39. Higginbotham, J. 1983. The logic of perceptual reports: an extensional alternative to situation semantics. Journal of Philosophy 80: 100 27. Mele, A and D. Robb. 1998. Rescuing Frankfurt-style cases. Philosophical Review 107: 97 112. Parsons, T. 1990. Events in the Semantics of English: a Study in Subatomic Semantics. Cambridge, Mass.: MIT Press. Ramsey, F. 1978. Facts and propositions. In his Foundations: Essays in Philosophy, Logic, Mathematics and Economics, ed. D. H. Mellor, 40 57. London: Routledge and Kegan Paul. Reichenbach, H. 1947. Elements of Symbolic Logic. New York: The Free Press. Van Inwagen, P. 1978. Ability and responsibility. Philosophical Review 87: 201 24. Van Inwagen, P. 1983. An Essay on Free Will. Oxford: Oxford University Press. Widerker, D. 1995. Libertarianism and Frankfurt s attack on the principle of alternative possibilities. Philosophical Review 104: 247 61. Wiggins, D. 1985. Verbs, adverbs and some other modes of grammatical combination. Proceedings of the Aristotelian Society 86: 273 304. Wiggins, D. 1991. Towards a reasonable libertarianism. In his Needs, Values, Truth, 269 300. Oxford: Blackwell. 7 Peter Van Inwagen (1978 and 1983, ch.5) is the best-known of these; he suggests that all of the following principles fail to succumb to Frankfurt-style counterexamples: PPA (Principle of Possible Action): A person is morally responsible for failing to perform a given act only if he could have performed that act. PPP1 (Principle of Possible Prevention 1): A person is morally responsible for a certain event-particular only if he could have prevented it. PPP2 (Principle of Possible Prevention 2) A person is morally responsible for a certain state of affairs only if (that state of affairs obtains and) he could have prevented it from obtaining.